[R-sig-ME] How to compute heritabilities on binomial response variables...

[Rudy Garel]

Hi everybody,

I am working on Atlantic salmon embryos and I wondered if anyone knows 
how to compute heritabilities on binomial response variables like survival?
Here is the model that I use to extract the variance components on a 
gaussian response variable (cf R book, p.640):

lmer(Bodylength~1+(1|Fem)+(1|factor(Mal))+(1|Fem:factor(Mal))+(1|factor(Inc)))

... where "Fem" is the female or mother effect, "Mal" the male or father 
effect and "Inc" the incubator effect.
And so I tried to extract the variance components on a binomial response 
variable drawing inspiration of this model as follows:

glmer(Survival~1+(1|Fem)+(1|factor(Mal))+(1|Fem:factor(Mal))+(1|factor(Inc)),family="binomial") 


The problem with this last model is that it does not give me the 
residual variance (which is normal with binomial data) and so I can not 
compute heritability.
Does anyone know an approach on R to compute heritabilities on binomial 
response variables?
Thanks in advance for paying attention to this mail.

Cheers,

-- 
Rudy Garel, PhD Student
Department of Ecology and Evolution, Biophore Building
University of Lausanne, CH-1015 Lausanne, Switzerland
Phone: +41 (0)21 692 41 76
http://www.unil.ch/dee/page90602.html

-------------- next part --------------
A non-text attachment was scrubbed...
Name: R book p640.pdf
Type: application/pdf
Size: 230985 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mixed-models/attachments/20121207/46b20c31/attachment-0001.pdf>